Abstract: In this paper, the prediction of acrylamide content in fried battered and breaded fish nuggets (BBFNs) was performed through combined response surface methodology (RSM) and back propagation artificial neural network (BP-ANN). RSM was utilized to collect the experimental data and an BP-ANN model was established to predict the changes of acrylamide content in BBFNs during deep-fat frying, which the ratio of xanthan gum to soybean fiber, drying time of BBFNs, soybean oil quality, frying temperature and time were considered as the input, as well as acrylamide content was regarded as the output. Furthermore, the training set was used for model fitting and the test set was used to evaluate the prediction ability of the model. The results showed that the acrylamide content in fried BBFNs was obviously affected by the ratio of xanthan gum to soybean fiber, drying time of BBFNs, frying temperature and time. However, the quality of soybean oil had no significant influence on the acrylamide content in fried BBFNs. R value (0.997) for correlation coefficient of the ANN model after training was presented, indicating that the model was notably fitted and had a good approximation ability. Moreover, the model had slight prediction errors for the new data with a maximum relative error (5.34%) and a minimum relative error (0.12%), suggesting accuracy prediction of the acrylamide content in fried BBFNs using BP-ANN.

Key words: battered and breaded fish nuggets, artificial neural network model, deep-fat frying, acrylamide content, prediction